A novel generative model for brain tumor detection using magnetic resonance imaging

Area of Science:

• Medical Imaging
• Artificial Intelligence
• Computational Biology

Background:

• Brain tumors are a significant global health concern requiring early and accurate diagnosis.
• Current diagnostic methods can be improved with advanced computational approaches for better patient outcomes.

Purpose of the Study:

• To develop and validate a novel AI-based method for precise brain tumor segmentation and classification.
• To enhance diagnostic accuracy and efficiency in brain tumor detection using deep learning and intelligent computational cells.

Main Methods:

• Utilized the "You Only Look Once" (Yolov8) deep learning framework for tumor region detection.
• Employed intelligent computational cells for fine-tuning and precise segmentation of tumor edges.
• Implemented a classification pipeline with grid search and data fusion across multiple datasets.
• Integrated Large Language Models (LLMs) for generative AI-based pre-diagnosis.

Main Results:

• Achieved over 98% accuracy for region detection and binary classification.
• Reached over 99% accuracy for brain tumor segmentation with segmentation times under 1 second.
• Demonstrated superior performance compared to state-of-the-art methods on the same dataset.
• Validated segmentation and classification through comparison with existing literature.

Conclusions:

• The proposed AI-driven Computer-Aided Diagnosis (CAD) method significantly improves brain tumor segmentation and classification accuracy.
• The integration of deep learning, digital image processing, data fusion, and LLMs offers a powerful tool for medical imaging diagnosis.
• This approach shows potential for aiding clinicians in early diagnosis, treatment monitoring, and predicting patient lifespan.